Guessing Machine

Guessing Machine Demo

This a Rails application that takes in a users height and weight to predict if they like Dogs or Cats. It uses ID3 (Iterative Dichotomiser 3) is an algorithm invented by Ross Quinlan used to generate a decision tree from a dataset.

Overview and Objectives

Overall approach as in any software and network application is the attempt to adhere to Roy Fielding’s REST Architecture where the goal it align with any number of his 7 principals around Performance, Scalability, Simplicity, Modifiability, Visibility, Portability, and Reliability where they best apply. Secondary objective was to create an easy to user interface where the participant would be provided with a simple to use interface that wouldn't be intimidating and games like.

• Performance: As the main consideration given that is a key part of the user experience, where interactions are app-like getting as close as possible to 60fps or less than 4 milliseconds for content loading 'time to interact and first meaningful paint'. So considering factors such as database performance, loading of assets and simplicity of the overall architecture.
• Scalability & Modifiability: Both from a business and developer perspective, the attempt was design the application to be self-scaling as more users engage with the application therefore more training data would be captured to provide greater guessing accuracy. Keeping in mind easy to read and understand code so any developer can quickly make improvements. Also making it easy to test and to have intraspection into how the application operates.
• Visibility & Reliability: Using comprehensive unit tests to the critical aspects of classification algorithm accuracy as well as performance. The unit test attempt to expose what is the threshold amount of training data needed for the highest correct responses.
• Portability: At minimum with the use of migration and seeders to interact with the test data for Reliability

In summary the attempt is to 'Fall into the Pit of Success' and avoid a 'Big Ball of Mud'

Installation and Dependencies

• Ruby version Ruby 2.6.1p33 and Rails 5.2.2.1

Installation

Install Ruby On Rails on Mac OS X 10.14 Mojave

System dependencies (Gems)

• sqlite3
• pg
• bootstrap
• chartjs-ror
• decisiontree

Decision Tree

A Ruby library which implements ID3 (information gain) algorithm for decision tree learning. Currently, continuous and discrete datasets can be learned.

Configuration

• To include main.js the following to config/initializers/assets.rb

Rails.application.config.assets.precompile += %w( main.js )

• Add the following to app/assets/javascripts/application.js

//= require jquery3
//= require popper
//= require bootstrap-sprockets
//= require Chart.bundle.min

• Database Starting Postgress

pg_ctl -D /usr/local/var/postgres -l /usr/local/var/postgres/server.log start

• Migration local development

rake db:drop db:create db:migrate db:seed RAILS_ENV=development

• Starting development server

rails s -e development

Heroku Deployment and setup

• Deploying to Heroku Commit and push as usual git commit... git push then push to heroku

git push heroku master

• Postgresql installation Configure database.yml with your database username and password

heroku addons:create heroku-postgresql

• Migration

heroku rake db:migrate db:seed

Testing

To models to use training and testing in various amounts and ratios

rails test:models -v RAILS_ENV=development

To test basic page functionality

rails test:controllers -v RAILS_ENV=test

CSV Data, SQL and ActiveRecord

Animals	Height	Weight
Dog	61.866	136.39
Cat	70.36	185.85
Dog	66.117	137.61

• CSV data loaded into sqlite3 for flexibility to sort, shuffle, and slice easily

testing = [
  ["Dog", 0.5787e2, 0.93534e2], ["Cat", 0.70974e2, 0.18498e3], ["Dog", 0.66021e2, 0.14166e3], ["Dog", 0.61964e2, 0.11283e3], ["Dog", 0.65003e2, 0.15606e3], ["Dog", 0.63723e2, 0.14245e3], ["Dog", 0.62361e2, 0.14356e3], ["Cat", 0.68792e2, 0.19768e3]
]

Default Settings

Settings	Value	Notes
Default seed training data	2000
Bias	Random	'Dog', 'Cat'
Real training data threshold	15	after the 15th human entered inputs the system stop using seed data from CSV
Real training data	2000

Performance Benchmarks

local testing

production testing

Unit testing on the imported sample data

$ rails test:models RAILS_ENV=development -v
Run options: --seed 18002

# Running:

Finished in 7.968875s, 1.7568 runs/s, 2952.7380 assertions/s.
14 runs, 23530 assertions, 0 failures, 0 errors, 0 skips